THE BUDDHIST FORUM, VOLUME V
A PHILOLOGICAL APPROACH TO BUDDHISM
The Bukkyō Dendō Kyōkai Lectures 1994
K.R. Norman
THE INSTITUTE OF BUDDHIST STUDIES, TRING, UK THE INSTITUTE OF BUDDHIST STUDIES, BERKELEY, USA 2012
First published by the School of Oriental and African Studies (University of London), 1997
© Online copyright 2012 belongs to: The Institute of Buddhist Studies, Tring, UK & The Institute of Buddhist Studies, Berkeley, USA
British Library Cataloguing in Publication Data
ISBN 0-7286-0276-8 ISSN 0959-0595
CONTENTS
The online pagination 2012 corresponds to the hard copy pagination 1997
Foreword...... vii
Abbreviations...... ix
Bibliography...... xi
I Buddhism and Philology...... 1
II Buddhism and its Origins...... 21
III Buddhism and Oral Tradition...... 41
IV Buddhism and Regional Dialects...... 59
V Buddhism and Writing...... 77
VI Buddhism and Sanskritisation...... 95
VII Buddhism and Aśoka...... 113
VIII Buddhism and Canonicity...... 131
IX Buddhism and the Commentarial Tradition...... 149
X Buddhism and Philology...... 167
Word Index...... 185
ABBREVIATIONS
AMg Ardha-Māgadhī AO Acta Orientalia B.C.E. Before Common Era BHS Buddhist Hybrid Sanskrit BHSD Buddhist Hybrid Sanskrit Dictionary BHSG Buddhist Hybrid Sanskrit Grammar BSO(A)S Bulletin of the School of Oriental (and African) Studies BSR Buddhist Studies Review BUp Bṛhad-Āraṇyaka Upaniṣad C.E. Common Era CP I–V K.R. Norman. Collected Papers, Volumes I–IV (1990–94) CPD A Critical Pāli Dictionary, Copenhagen CUP Cambridge University Press DEDR T. Burrow & M.B. Emeneau, Dravidian Etymological Dictionary (Revised edition, 1984) DPPN G.P. Malalasekera, Dictionary of Pāli Proper Names (1937–38) EV I, II K.R. Norman, Elders’ Verses I (1969), II (1971) GD I, II K.R. Norman, The Group of Discourses I (1984), II (1992) IIJ Indo-Iranian Journal IT Indologica Taurinensia JA Journal Asiatique JBuRS Journal of the Burma Research Society JOI(B) Journal of the Oriental Institute (Baroda) JPTS Journal of the Pali Texts Society JRAS Journal of the Royal Asiatic Society m.c. metri causa MIA Middle Indo-Aryan MRI Minor Readings and Illustrator
MW M. Monier-Williams, Sanskrit-English Dictionary (1899) OUP Oxford University Press PDhp Patna Dharmapada PE Pillar Edict PED The PTS’s Pali-English Dictionary Pkt Prakrit PTS Pali Text Society RE Rock Edict SBE Sacred Books of the East Skt Sanskrit SOAS School of Oriental and African Studies (London) SWTF Goerg von Simson (ed.), Sanskrit-Wörterbuch der buddhistischen Texte aus den Turfan-Funden (1973–) Überbl O. von Hinüber, Das ältere Mittelindisch im Überblick (1986) WZKSO Wiener Zeitschrift für die Kunde Südasiens
Abbreviations of the titles of Pāli texts are those adopted by the CPD. Editions are those of the PTS, unless otherwise stated.!
IX Buddhism and the Commentarial Tradition
Commentaries were probably needed from the very beginning of Buddhism. We can assume that the earliest form of commentary was simply the explanation of one word by another, and we need not doubt that the Buddha on occasion replaced words by synonyms, to make his teachings more easily understood. We find some sort of evidence of this in the words of the Buddha himself, where he gives, for example, seven different words all meaning “bowl”,1 which he says are “country language” (janapadanirutti). This is presumably something which he found from experience, i.e. as he moved from one place to another, he found that the word used for a particular object in one area was quite different from that used in another area for one and the same thing. Anyone who has ever looked at Sir George Grierson’s Bihar Peasant Life,2 will know exactly the problems which the Buddha faced. Grierson moved around the area for which he had responsibility as a civil servant, recording the wide range of different names which the villagers of Bihar used for all the items of everyday domestic and village life. The Buddha must have found a situation which was very similar to this 2 ½ millennia ago.
The earliest extensive portion of Theravādin commentary material we have is that found in the Vinaya-piṭaka. We know that, at the suggestion of King Seniya Bimbisāra of Magadha,3 the Buddha instituted a fortnightly recitation of the Pātimokkha (following the practice of other sects which recited texts). We are therefore certain that the Pātimokkha existed from the very early days of Buddhism. The Pātimokkha does not have a separate existence in the Theravādin tradition, but is embedded in the Vinaya-piṭaka. If it is the earliest text for which we have direct evidence of recitation, it is not surprising if it was the first text that needed explanation. And so we find a simple type of commentary upon it, whereby words are for the most part explained by synonyms or clarification of terms, e.g. the first word of the first pārājika rule is yo “whoever”, and it is explained: “‘Whoever’ means he who on account of his relations, his social standing, his name, his clan, his morals, his dwelling, his field of activity, an
1 M III 234, 34 foll. 2 Grierson, 1926. 3 See Vin I 101, 18 foll.
149 elder, or a novice, or one of middle standing, this is ‘whoever’”.4 The nature of that commentary is made clear by its name: pada-bhājaniya “analysis of words”. The commentary is so early that, like the Pātimokkha itself, it is embedded in the Vinaya- piṭaka after each of the Pātimokkha rules, and therefore the text and the commentary have been handed down together.
The earliest Theravādin commentarial text is one which is actually given separate canonical status. It is the Niddesa, a commentary upon two vaggas and one sutta of the Sutta-nipāta. Here again we find a great deal of explanation by means of synonyms. We also find another characteristic feature of this type of commentary. Whenever a particular word occurs in the text, the same explanation is given verbatim, e.g. whenever there is a reference to kappa “figment”, or a verb based upon the root kapp-, we are told that there are two types of kappa, i.e. taṇhā-kappa and diṭṭhi-kappa, and the identical explanation is given, even if the words recur in successive verses. In the third lecture, I pointed out that this kind of fixed repetition is typical of the oral nature of an early text, giving support, if it were needed, for the assumption that the Niddesa is very old and was composed well before the writing down of the canon.
Even the early commentarial passages in the canon itself are not merely lists of synonyms. Commentaries have two functions, although not all commentaries perform both of them all the time: they explain the meaning of the words and they explain the meaning of the phrase or sentence in which the words occur. In the Araṇavibhaṅgasutta “The discourse on the undefiled” of the Majjhima-nikāya,5 the araṇa “undefiled person” is defined in one paragraph by the Buddha. The remainder of the sutta is devoted to explaining the meaning of each sentence of that paragraph. Similarly, the Niddesa is not merely a list of synonyms. It also includes exegetical passages, giving us some idea of what the early commentarial tradition thought was the meaning of the relevant portions of the Sutta-nipāta. It is, however, not a organic structure of exegesis, but a series of disconnected phrases, which serve as explanations of the individual words, not in the particular context of the Sutta-nipāta, but in any setting.
The chronicles tell us that, at the time of Aśoka, Mahinda took Buddhism to Sri Lanka. 6 The story, as related in the Dīpavaṃsa, the Samantapāsādikā, and the Mahāvaṃsa, tells how King Tissa met Mahinda, exchanged greetings with him and had a discussion. All three sources agree that the first sermon which Mahinda preached to the king was the Cūḷahatthipadūpamasuttanta “The short
4 Vin III 23, 37 foll. 5 M III 230–37. 6 Dīp 12.9; Mhv 13.21.
150 sermon on the elephant’s footprint”.7 The Mahāvaṃsa8 states that Mahinda preached the true doctrine in two places “in the speech of the island” (dvīsu ṭhānesu dhammaṃ bhāsitvā dīpabhāsāya), implying that in the other places he did not “translate”.9 The Dīpavaṃsa and the Samantapāsādikā make no mention of this “translation” process, and although there are instances of the Mahāvaṃsa including authentic historical material which was either unknown to, or consciously omitted by, earlier chroniclers, it is possible that the need for “translation” was deduced by the author of the Mahāvaṃsa from the statement repeated at the beginning of Buddhaghosa’s commentaries, that they were translations into Pāli of the commentaries which were brought to Sri Lanka by Mahinda and put into Sinhalese for the benefit of the islanders.10 The author of the Mahāvaṃsa may well have thought that if Mahinda needed to “translate” the commentaries then he must have “translated” the canonical sermons too. The fact that the “translation” process is said to have happened in only two places suggests, however, that the statement is based upon tradition. If it is true, we may deduce that in some places Mahinda preached the suttas in their canonical form, whereas in two places he added an exegesis, in Sinhalese Prakrit.
There is no evidence about the language of the commentaries which Mahinda is said to have brought with him to Sri Lanka. They could have been in the same language as the canonical texts, whatever that was, or they might have been in the language of the missionaries, in so far as they may have spoken a different Prakrit. We have no evidence earlier than Buddhaghosa’s statement which I have just mentioned,11 that Mahinda did bring commentaries with him, but we have no reason to doubt that commentaries were brought from India either by Mahinda or other missionaries coming after him. The commentaries seem to have contained much information about India, and particularly North India, which could only have been brought from there. As we shall see, the Pāli commentaries sometimes employ a form of words in their explanations which is closer to the versions found in Sanskrit or the Gāndhārī Prakrit than the Pāli version which is being commented upon,12 and this can only be explained by assuming that the Pāli commentary and the Sanskrit and Gāndhārī versions go back to a common source, which must have been in North India.
7 = M I 175–84. 8 Mhv 14.65. 9 I use “translate” to mean the change from one dialect of Middle Indo-Aryan to another, however closely related. 10 Sīhaḷadīpaṃ pana ābhatā ’tha vasinā Mahā-Mahindena ṭhapitā Sīhaḷabhāsāya dīpavāsīnam atthāya (Ps I 1, 23*–24* and Lamotte, 1988, 557). 11 Ps I 1, 21* foll. 12 See Brough, 1962, 226.
151 There is evidence for dialect differences in the early commentarial material. We know from the Aśokan inscriptions that in the third century B.C.E. there was a dialect variation in the treatment of consonant groups. In the East the dialects tended to resolve consonant groups by inserting an epenthetic (svarabhakti) vowel. In the West such consonant groups were simplified by assimilation. We occasionally find that a word is found in the Pāli canon which shows one regional variation, while the explanation of the word which is preserved in the commentaries shows a cognate form from another dialect, e.g. we find in the Dhammapada13 the phrase vissaṃ dhammaṃ samādāya, usually translated “having adopted the whole law”, assuming that vissa is from Sanskrit viśva “all”.14 The commentary, however, explains this as “vissa means taking up a dhamma that is uneven (visama) or a dhamma that is foul-smelling, concerned with physical activities, etc.”.15 The alternative explanation signified by “or” (vā) shows that the commentator was uncertain about the meaning of the word, but nevertheless the tradition of the Sinhalese aṭṭhakathā he was following had preserved an alternative dialect form.
One Sanskrit version of the verse has veśma-dharma16 and it is probable that the Gāndhārī17 version also has veśma-. The meaning therefore is “domestic manner (of life)”.18 In the Pāli version vissa stands for věssa, where -ss- results from the assimilation of -śm- > -ss-. The first explanation in the commentary shows the alternative development of an epenthetic vowel being inserted, i.e. veśma > *viśma > *visma > visama, which the commentator took to be the equivalent of Sanskrit viṣama “uneven”. Another Sanskrit version of the same verse actually has viṣama,19 which is clearly a backformation from the same tradition which preserved the Pāli commentarial reading visama. We may presume that the verse was uttered at least twice, once in an area where consonant groups were assimilated, and once where they were resolved. These explanations in different dialects must therefore antedate the introduction of the canon into Sri Lanka, and they show that the dialect variation already existed at an early date, perhaps at the time of the Buddha.
Other explanations which depend upon non-Pāli and non-Sinhalese Prakrit forms must also antedate the introduction of the canon into Sri Lanka, e.g. the
13 Dhp 266. 14 It is so taken by PED, s.v. vissa1. 15 vissan ti, visamaṃ vissa-gandhaṃ vā kāya-kammādikaṃ dhammaṃ samādāya, Dhp–a III 393, 2–4. 16 Udāna-v 32.18. 17 GDhp 67. 18 Brough, 1962, 266. 19 Mvu III 422, 13*.
152 explanation of coḷa20 “top-knot” by cora “thief” must depend upon a source which, like Māgadhī, confused l (ḷ) and r. Similarly, abhibhāsana21 “illuminating” is explained as tosana “delighting”,22 but this is really the explanation of abhihāsana,23 and shows that the explanation must have been made in a dialect where both abhibhāsana and abhihāsana appeared as abhihāsana.
The Critical Pāli Dictionary24 refers to nominative singular forms in -e in two commentarial passages25 and claims that these are from the
Childers was of the opinion28 that the commentaries brought by Mahinda were in Pāli. If Childers was right, we have to assume a progression from Pāli to Sinhalese Prakrit and back to Pāli again. While this is not impossible, it does raise the question of why the Pāli of the canon was sacrosanct, while the Pāli of the commentary could be translated. It also raises the question of the nature of Pāli. If it means the language of the Theravādin canon, then it is unlikely that at the time of Aśoka, when Mahinda came to Sri Lanka, the language of that canon was already a western Prakrit with a certain amount of Sanskritisation introduced into it. I think that the Sanskritisation came later, perhaps because of the rise of the Mahāyāna. It would seem more likely that the commentaries already represented a heterogeneous mass of material, in various dialects, and
20 Th 170. 21 Th 613. 22 abhibhāsanan ti tosanaṃ appaṭisāra-hetutāya cittassādhippamodanato, Th-a II 260, 2–3. 23 abhihāsana < *abhihassana < abhiharṣaṇa. 24 CPD, s.v. avitakka. 25 ese eke ekaṭṭhe same samabhāge tajjāte tañ ñevā ti, Mp I 71,13; ese eke ekaṭṭhe same sabhāge tajjāte tañ ñevā ti, II 273, 16–17. 26 ese se ekaṭṭhe same samabhāge tajjāte ti, Kv 26, 20–21 et passim. 27 See Norman, 1979A, 279–87 (281) (= CP II 59–70 [61]). 28 Childers, 1875, x.
153 probably including comments on readings which differed from those in the canon as established in Sri Lanka.
In the fifth lecture, on Buddhism and writing, I quoted the passage in the Dīpavaṃsa which stated that prior to the reign of Vaṭṭagāmini Abhaya the tipiṭaka and the aṭṭhakathā were transmitted orally, but in his reign they were written down. We tend to concentrate on the fact that it was the tipiṭaka which was written down, without thinking much about the commentaries because, to us, the word “commentaries” implies the works composed, in particular, by Buddhaghosa and Dhammapāla in the fifth and sixth centuries C.E. respectively. The reference in the Dīpavaṃsa must, however, be to whatever commentarial material was available in the first century B.C.E., and which had hitherto been transmitted orally in conjunction with the canonical texts to which it referred.
The fact that the commentarial material was already of a disparate nature would probably have led to an attempt to impose homogeneity upon it, and also to make it more intelligible to the Sinhalese bhikkhus by translating it into the vernacular language. Because it was in the vernacular it would have been easy for additions to be made to it. That there was a continuing commentarial tradition in Sri Lanka itself is shown by the fact that Buddhaghosa quotes (from the Sinhalese commentaries he was using) the names of individual Sinhalese theras whose views he is accepting or rejecting.
According to the Cūlavaṃsa, there were no commentaries available in India in the fifth century C.E., and so the thera Revatta suggested to Buddhaghosa that he went to Sri Lanka and translated the Sīhaḷa commentaries into Māgadhī.29 At the beginning of the Samantapāsādikā, Buddhaghosa states that his work will be based on the Mahā- aṭṭhakathā and the Mahāpaccariya, while also taking into account such commentaries as the Kurundi. He also quotes from the Andhaka and the Saṅkhepa, and the Paccarī, but it is not clear whether this last named is the same as the Mahāpaccariya.
There has been much speculation about the Andhaka-aṭṭhakathā;. It is said30 that it was handed down at Kāñcipura in South India. It is further stated31 that it was very likely written in the Andhaka language. Probability then becomes certainty, and we than find statements such as “the references relate to … the Sinhalese and some of the Dravidian commentaries”,32 and “Buddhaghosa drew his material not only from Sinhalese and Dravidian but also from … Pāli”.33 Comparable statements are made about Dhammapāla: “Dhammapāla … very
29 Mhv 37.230. 30 C.A.F. Rhys Davids, 1974, xxviii. 31 Adikaram, 1953, 12. 32 Adikaram, 1953, 14. 33 Adikaram, 1953, 16.
154 probably was a Dravidian by birth. It is also likely that he made use of Dravidian commentaries”.34
There was, in all probability, some connection between the Andhaka commentary and either the Andhaka country or the Andhaka sect, and it is very likely that the sect was so called because it came from the country. Probably the Andhaka commentary came from the Andhaka country originally, for Buddhaghosa quotes it 35 as referring to conditions in that country. It must, however, like the other commentaries, have had a core of material which came from North India, for Buddhaghosa quotes the Andhaka- aṭṭhakathā when he is talking about the Magadha-nāḷī36 “a particular measure” but quotes the Mahā-aṭṭhakathā for the Damiḷa-nāḷī in the same passage. The fact that the Andhaka commentary is usually quoted only to be refuted37 tends to support the view that the basis, at least, of the Andhaka commentary belonged to the Andhaka sect.
About the language of the Andhaka commentary nothing can be said definitely, but it would seem clear that by Buddaghosa’s time it was no longer available in South India. The statement that no commentary was available in India,38 which is given as the reason for Buddhaghosa to go to Sri Lanka, was not likely to have been included in the Sinhalese commentaries upon which the Mahāvaṃsa was based while there were South Indians such as Dhammapāla available to refute it, if it was false. It is in any case clear, from Buddhaghosa’s statement,39 that the view of the thera Mahāsumma was regarded as authoritative in the Andhaka-aṭṭhakathā, that the commentary must have been introduced to Sri Lanka some time before Buddhaghosa, for Mahāsumma is a Sinhalese thera who is datable to the first century C.E. Since, however, Buddhaghosa stated plainly that he translated Sinhalese commentaries into Māgadhī, we can be fairly certain that if the Andhaka commentary was originally composed in a Dravidian language, it had already been translated into Sinhalese Prakrit by Buddhaghosa’s time.
It is usually stated40 that no material was added to the Sinhalese commentaries after the first century C.E., but a very careful study41 of the individuals mentioned in the commentaries has shown that King Mahāsena is mentioned by name in the Samantapāsādikā.42 Since this king is datable to 276–303 C.E., this
34 Adikaram, 1953, 9. 35 Sp 747, 23. 36 Sp 702, 23–27. 37 e.g. Sp 697, 1. 38 Mhv 37.227. 39 Sp 646,11. 40 e.g. by Adikaram, 1953. 41 Mori, 1988, 119–67 (143). 42 Sp 519, 26.
155 shows that additions to the Sīhaḷa-aṭṭhakathās continued to be made until the very end of the third century or the beginning of the fourth.
The author of the commentary upon the Sutta-nipāta notes that the Sinhalese commentary did not comment upon two verses in the Sutta-nipāta,43 and he therefore excludes those verses from the number of verses in the original sutta.44 This statement has been taken45 as pointing with more or less certainty to an addition made to the canon in Sri Lanka. The verses could equally well, however, have belonged to a different recension from that being used by the Mahā-aṭṭhakathā, and been either unknown to the author of the Sīhaḷa commentary or ignored as being less authentic. It is clear that different recensions of some texts did find their way to Sri Lanka. Buddhaghosa and the other commentators frequently record the existence of variant readings, and it is questionable whether these would all have been in one and the same version of the canon. It is interesting that the commentator on the Sutta-nipāta did not invent a commentary upon these verses. It would seem that the commentary was, as far as he was concerned, closed.
If the Buddha himself explained X by Y in one place, but Y by X in another, then we can see that X and Y were both Buddhavacana, and this situation might lead to a different sort of commentarial tradition, whereby the commentarial explanation in one tradition was the canonical reading in another tradition, and vice versa. I have just mentioned the comparable situation with regard to vissa, veśma and visama.
It has been shown46 that in a number of cases other (non-Pāli) traditions preserve as canonical forms which are akin to those in the Pāli commentary, showing that those traditions thought that they were Buddhavacana. For example, where a verse in the Dhammapada 47 includes the words adhisessati, chuddho, and apeta-viññāṇo, the Gāndhārī Dharmapada equivalent48 has three different words in their place (vari śa’iṣadi, tuchu, and avakada-viñaṇa). The glosses in the Pāli commentary, however, are the equivalents of the Gāndhārī forms (upari sayissati, tuccha, and apagata-viññāṇa, Dhp-a I 320–21), which clearly shows acquaintance with the same tradition as underlies the Gāndhārī version. A possible explanation of this is that the Buddha uttered the same sermon on more than one occasion, sometimes making changes, as the occasion
43 Sn 677–78. 44 avasāne gāthādvayaṃ eva pana Mahā-aṭṭhakathāyaṃ vinicchitapāṭhe n’ atthi, tenāvocumha “vīsatigāthāsū” ti, Pj II 477, 13–14. This seems to be the only statement of this kind in the cty. 45 Adikaram, 1953, 12. 46 Brough, 1962, 192. 47 Dhp 41. 48 GDhp 153.
156 or locality demanded. When tradition preserved more than one version, it sometimes kept one as an explanation of the other.
Another example of this parallelism between canonical and commentarial traditions, which seems to cross the boundaries of sect, can be seen in Buddhaghosa’s commentary on the incident where bhikkhus ask for permission to change the Buddha’s words chandaso. He glosses: “Let us translate chandaso means, let us translate into Sanskrit (or into the refined language) as (we translate) the Veda”.49 Here “into the refined language” (sakkata-bhāsāya) seems to correspond to phrases in two of the Chinese versions of the same incident. The Dharmaguptaka version includes the request in the form “in accordance with the fine language (or perhaps texts) of the world”, while in the Vinaya-mātṛka version the Buddha states that his doctrine “is not concerned with beautiful language”. It seems very likely that both these canonical phrases are based upon originals which included the equivalent of saṃskṛta or satkṛta.50
It is an interesting fact that in his commentary on the heretics’ views which are quoted in the Sāmaññaphala-sutta,51 Buddhaghosa seems to include phrases which are akin to those in the Tibetan version of the Pravrajyā-vastu.52 He glosses: “cooking means beating others with a stick”,53 which seems to be a combination of two phrases in the Tibetan: ‘Who grills and lets grill, who beats and lets beat’.54 At the end of his explanation of Ajita’s doctrines, he comments: “fools give, the wise take”.55 The Tibetan version states: ‘Thus (only) the fool accords instruction, the sage receives instruction’.56
These parallels suggest that the commentarial and canonical traditions which underlie them are equally old, and the material upon which they are based must pre-date the separation of the sects. It seems probable that some of this material, since it is accepted as canonical by some sects, must go back to the earliest days of Buddhism, perhaps to the time of the Buddha himself. A close comparison of the Theravādin commentaries with non-Theravādin canonical texts might well bring to light other parallels of this nature. This might help us to date the various strata of the commentarial tradition, which would, in turn, enable us to estimate its value more accurately.
49 chandaso āropemā ti, vedaṃ viya sakkata-bhāsāya vācanāmaggaṃ āropema (Sp 1214, 16–17). 50 See Brough, 1980, 35–42 (39). 51 D I 52 foll. 52 Vogel, 1970. 53 pacato ti, pare daṇḍena pīḷentassa (Sv 159, 16–17). 54 Vogel, 1970, 26. 55 bālā denti, paṇḍitā gaṇhantī ti dasseti (Sv 166, 16). 56 Vogel, 1970, 22.
157 Sometimes there is mention of an event in a commentary, with no reference at all to it in the canon, although it appears in the canon of another school. As I mentioned in the eighth lecture, a problem arises from the fact that some of the best known stories in Buddhism are known in Pāli only in the commentaries, e.g. the story of the kṣatriya maiden Kisāgotamī.57
In some cases, portions of Buddhist history seem not to have reached Sri Lanka at all, e.g. the early history of Devadatta’s hostility to the Buddha, and the reason why the Buddha called him “lick-spittle”,58 kheḷāsika.59 Buddhaghosa does not know the reason, and the incorrect explanation he gives 60 is followed by modern translators and commentators, so that, for example, Miss Horner translates it “to be vomitted like spittle”. Some of the Chinese versions of the Vinaya, however, do know the correct meaning of kheḷāsika, and tell a story, perhaps invented to explain the word, of Devadatta eating Ajātaśatru’s spittle.
Sometimes the commentators give explanations based upon individual syllables of words, e.g. bhikkhu: saṃsāre bhayaṃ ikkhanatāya, “because of seeing danger in saṃsāra” Vism 16, 21; bhagavat: vanta-gamano bhavesu, “he has rejected going in existences” Pj I 107, 26*; King Pasenadi is so called because: paccāmittaṃ parasenaṃ jinātī, “he conquers the hostile army of another” Ud-a 104, 27. Since the Sanskrit form of his name was Prasenajit [Pkt Paseṇai], the commentarial explanation is correct to include “he conquers”, and this makes it clear that when this explanation was composed (which might have been long before Dhammapāla) the name still had the spelling with -j-, or the tradition at least knew that the correct form was -j-.61
The fact that the commentary explains the syllable -di by jināti, shows that the text and the commentary were transmitted separately,62 with neither having an effect upon the other. The reason for the -d- of the text form is not clear. It is claimed by some as a Sinhalesism, because j > d in Sinhalese. If this were the explanation, I should have expected the commentary to show the change, not the text. The disagreement between the canonical text and the commentary is probably due to the fact that they were in the keeping of different groups, i.e. the bhāṇakas responsible for the canonical text were not also responsible for the safe keeping of the commentary upon that text.
Other explanations are given in accordance with the type of etymologising in Sanskrit literature known as nirukti or nirvacana, which was taken over into
57 See Collins, 1990, 89–126 (117, note 55). 58 At Vin II 188, 37 and Dhp-a I 140, 1. See Lamotte, 1970, 107–15. 59 See Norman, 1979B, 325–31 (329, note 23) (= CP II, 187–93 [191, note 2]). 60 Sp 1275,17. 61 Aggavaṃsa has a rule to deal with this: Sadd sūtra 88 (622,24): do jassa: Pasenadi. 62 For similar separate transmission of text and commentary, see von Hinüber, 1981, 74–86 (78).
158 Buddhist literature. It often resembles folk etymology in its mode of operation, but this is to misunderstand its purpose. Quite frequently words are explained by means of others which are similar in appearance but are, in fact, in no way related to them. Such etymologies are found in the canonical texts from an early period, and are then taken over into the commentarial tradition, e.g. brāhmaṇa, based upon a root brah- meaning “to be strong”, is explained in a Dhammapada verse by means of a different root brah- “to remove”. 63 A brāhmaṇa was a remover—of his evil deeds, of course. Similarly a samaṇa is so called because he has put to rest his evil deeds, explaining sam- (Skt śram- “to strive”) by sam- (Skt śam- “to be quiet”).64 The word rājā “king” is explained in the Dīgha-nikāya by “he pleases others by righteousness”, explaining rāj- “to rule” by means of rañj- “to be delighted”.65 The fact that a comparable etymology is given, or alluded to, several times in the Mahābhārata and Rāmāyaṇa66 shows that this is not a Buddhist invention.
Some of these explanations seem ludicrous in western eyes, and this has led to some modern commentators trying to see the joking nature of the Buddha at work in them. It has been claimed that the Buddha is here parodying the etymologies (nirukti) found in brahmanical texts.67 I cannot rule out the possibility of the Buddha joking, although it is noteworthy that joking is not a well-known or highly approved Buddhist pastime, but I can see no evidence that these etymologies were really intended as jokes. Certainly in a series of etymologies of this type in the Sabhiya-sutta of the Sutta-nipāta, dealing with the words kusala, paṇḍita, muni, etc., no joke seems to be intended, and I find it hard to understand why we should think that some etymologies are meant to be humorous, and others not.
I am reminded of talking to various people about Sanskrit, and being asked by one if Sans-script was so called because it had no script and could not be written down, and by another if Sand-script was so called because the early Indians had no writing materials and had to scratch the characters in the sand. These folk-etymologies seemed comical to me, but both questioners were very serious about them.
These nirvacana etymologies are sometimes of value (beside the etymological theorising they contain) because they give information about the dialect(s) in which they were first given, and such information about the earlier history of the
63 bāhita-pāpo ti brāhmaṇo, Dhp 388. 64 samitattā hi pāpānaṃ samaṇo ti pavuccati, Dhp 265. 65 dhammena pare rañjeti, D III 93,14. See Gombrich, 1992, 159–78 (174). 66 See Hara, 1969, 489–99. 67 See Gombrich, 1992, 174.
159 text may not be available in other ways.68 For example, to explain brāhmaṇa by using the form bāhita shows that in the dialect in which the etymology was invented the word brāhmaṇa had the form *bāhaṇa, while some of the etymologies in the Sabhiya-sutta make sense only if we assume that they were first composed in a dialect in which -t- and -j- became -y-.
The fact that, despite the displacement of pādas, etc., which makes the etymologies in the Sabhiya-sutta very hard to understand, the commentary still contains the correct traditional explanation of some of the words included in the sutta, e.g. kusala and ājāniya, shows that the sutta tradition, which is very corrupt, and the commentary tradition, were quite separate, as in the case of the name Pasenadi which I have just mentioned, and the commentary went on transmitting the correct explanations even though the text which was being commented upon no longer had the correct readings.69
The growth of homonyms in MIA often meant that two or more explanations were possible for a word, depending upon which etymology was being considered. This sometimes enables a Pāli commentary to give a richer and fuller exegesis than would be possible in Sanskrit, where a redactor had to decided upon one form or another. The BHS equivalent of sammappadhāna “right effort” is samyakprahāṇa “right abandoning”. I mentioned in the sixth lecture the suggestion70 that the Sanskrit version is an incorrect backformation from an MIA form sammappahāna which might have either meaning. This view is supported by the fact that Buddhaghosa’s explanation of the term actually includes the idea of abandoning,71 and this suggests that the commentarial material was first composed in a dialect where the form of the word was sammappahāna.
Where the commentators no longer knew the meanings of words, they sometimes had to deduce them by devising etymologies which we can see are incorrect. In Pāli the consonant cluster in the word -jña “knowing” becomes -ññ-. The consonant cluster -ny- also becomes -ññ-. When dealing with the word vadaññu, which is derived from Skt vadānya “bountiful, liberal, a munificent giver” or “eloquent, speaking kindly or agreeably, affable”,72 the commentarial tradition wrongly took the final element -ññu to be from -jña, and the word is explained as vacana-vidu,73 i.e. understanding vadaññu as the equivalent of vada-jña “knowing the utterance (of the Buddha)”.
68 See Norman, 1980B, 173–84 (177–78) (= CP II, 148–64 [154–55]). 69 See Norman, 1980B, 173–84 (178) = (CP II, 148–64 [155–56]). 70 See Gethin, 1992, 70. 71 Gethin, 1992, 71. 72 See MW, s.v. vadānya. 73 P II 415 (ad Sn 487).
160 Another development from Sanskrit -jña is Pāli -jina, with an epenthetic (svarabhakti) vowel evolved between -j- and -ñ-. The resultant form is identical with the word jina “conqueror”, and the commentaries consequently explain compounds ending in -jina as “conqueror”, e.g. magga-jina74 “conqueror (of the defilements) by means of the road”,75 whereas we are probably to see its meaning as “knowing the road”. The same type of explanation is given for khettajina “conquering the field”. Here we can see that -jina is certainly from -jña, because there is a Sanskrit equivalent kṣetra-jña “knowing the field”.76
We are inclined, in western philology, to believe that there is only one correct answer to a question of etymology. In India, however, there was a custom of seeing more than one meaning in any word or phrase—the so-called śleṣa. So, instead of saying the meaning is either this or that, as we would do, commentators very often say that the meaning is this and that. Sometimes the meanings they give include what we would regard as the correct etymology, but sometimes they are all, from our point of view, incorrect. These are not, however, intended as western-style etymologies, but they have rather a religious, or even mystical, purpose.
So the word bhagavat is explained with the words: “The reverend one (guru) has blessings (bhagī), is a frequenter (bhajī), a partaker (bhāgī), a possessor of what has been analysed (vibhattavā), has caused abolishing (akāsi bhaggaṃ), is fortunate (bhāgyavā), has well maintained himself in being (subhāvitattano) in many ways, has gone to the end of being (bhav-anta-go), thus he is called bhagavat”.77 These multiple explanations would not be regarded in the tradition as specifying the meaning of the word bhagavat, but as attributes or epithets of the bhagavat himself, and would form the basis of a sort of litany “the bhagavat has blessings, etc.”, i.e. they would be taken as facts, not as explanations of the term bhagavat.
Similarly with arahat, which is explained: “One is an arahat because one’s defilement are far from one (āra-kilesa)”.78 “The bhagavat is arahanta because of remoteness (āraka), because of the destruction of his enemies (ari) and of the spokes (ara = rāga, etc.), because of his being worthy of requisites, and because of the absence of secret ill-doing (pāpakaraṇe rahābhāvā).79 The inclusion of an
74 Sn 84 75 Pj II 162: maggajino ti maggena sabbakilese vijitāvī. 76 See Dhadphale, 1980, 205, and Norman, 1980B, 173–84 (183, note 6) (= CP II, 148–61 [149, note 3]). 77 Pj I 107,19 foll.; translated at MRI, 116. 78 Ps I 42,19, quoting M I 280. 79 ārakattā arīnaṃ arānañ ca hatattā paccayādīnaṃ arahattā pāpa-karaṇe rahā-bhāvā ti imehi tāva karaṇehi so bhagavaṃ arahan ti veditabbo, Sv 146,10–12; cf. Vism 198,11 foll. cf. ārakattā arahā hoti, A IV 145, 2; kilesārīnaṃ hatattā arahatā (instr.sg.), Mp V 84, 22.
161 explanation based on ari is interesting because the Jain form is ari-hant, i.e. with an svarabhakti vowel -i- instead of -a-.
A number of non-Buddhist terms were taken over and applied in a Buddhist sense, e.g. the Buddha is called a yakṣa and a nāga. Since it was thought inappropriate to call the Buddha by the name of a minor supernormal being, i.e. a “snake-god”, other explanations were invented. One is a nāga because one does not perform sin (na + āgu), or one will not go to rebirth again (na + gam “to go”), or will not come back again (na + āgam “to come”).
Sometimes the explanations which the commentators give include forms which suggest that they were first given in some other dialect. So, among the explanations of tathāgata, we find some based upon forms with gada instead of gata, i.e. they are taken from, or through, a dialect which voiced -t- to -d-. Tathāgata is explained as: “He is thus gone (tathā gata), thus come (tathā āgata), he has thus known (tathā ājānana [verbs meaning ‘go’ also mean ‘know’]), has uttered what is real (tathā gadanato)”.80 We are also told that just as a physician overcomes snake-bite by an antidote (agada), so the tathāgata overcomes the world with the antidote of truth.81
Commentators not always well informed, and we can see that they or the sources they were following were sometimes mistaken. For example, in the commentary on the Sutta-nipāta,82 Ādiccabandhu is said to be the name of a pratyeka-buddha, presumably because it occurs in the verses ascribed to the pratyeka-buddhas. The word is, however, used so commonly as an epithet of the Buddha that I find it hard to believe that it does not apply to him in this verse.83
Ignorance about rare grammatical forms also led commentators astray. The word phalesin is a future active participle,84 and means “about to fruit”. It occurs in a verse in the Theragāthā85 where the trees are described as being in blossom and about to fruit. The commentator, unacquainted with the form, analyses phalesin as coming from phala “fruit” and esin “seeking”, and describes the trees as “seeking their fruit”. The word occurs again in a verse86 where a foolish action is likened to a man who cuts a tree down just when it is about to fruit. Here the error in the translation “a man seeking fruit cuts a tree down” is masked by the fact that this too would be a foolish action.
80 MRI 217 foll.; cf. Ud-a 131, 15. 81 See also Ud-a 132, 9 for agada/āgada. 82 Pj II 105, 27. 83 Sn 54. 84 See Geiger, 1994, § 193A. 85 aṅgārino dāni dumā bhadante phalesino chadanaṃ vippahāya, Th 527ab. 86 yathā phalesī mūle taruṃ chettu tam eva icchasi, Th 1121ab.
162 Although some of the commentarial material must go back to very early times in Buddhism, perhaps to the time of the Buddha himself, nevertheless, as we know well from other religions, all texts are interpreted in the light of the age and culture of the reader, which may differ considerably from the age and culture of the original composer of the text. We have, therefore, to decide how we should interpret a canonical text. Do we interpret it in the light of our knowledge of the culture of the time, the culture of the first commentary upon it, the culture of the second commentary on it, the culture of present- day Buddhists in South or South-east Asia, or our own western culture?
An examination of the Niddesa and the explanations it gives enables us to see that the composer of that text was already interpreting the ideas of the Sutta-nipāta in a different way from the original hearers of the text, as far as we can judge.87 We also have a commentary upon the Sutta-nipāta attributed to Buddhaghosa, and he too interprets the Sutta-nipāta in the light of his own culture, or at least the culture of those who wrote the material in the Mahāvihāra which he used in his work. Again there are differences of interpretation, as the commentaries tried to interpret things in their own way, although, of course, this does not necessarily mean that their interpretation differed from that of the Buddha’s contemporary audiences.
The way in which interpretations changed can be seen easily from an examination of the meaning of the phrase tasā vā thāvarā vā.88 If we base our translation on the meaning of the related Sanskrit words, then Sanskrit trasa means “moving” = “the collective body of moving or living beings” (as opposed to sthāvara),89 and Sanskrit sthāvara means “standing still, not moving, fixed, stable, immovable”.90 The phrase, therefore, means “(all creatures) moving or unmoving”. By the time the Pāli commentary was written, however, the interpretation of the words had been changed. The commentator states:91 “tasa is a synonym for those who have cravings or fears: thāvara means that they are standing still. This is a synonym for those who are rid of their cravings, i.e. arahats”. To the commentator, therefore, the phrase had taken on a Buddhist flavour: “Ordinary individuals who still had cravings, and arahats who were freed from them”. This explains why Miss Horner translated the compound tasa-thāvara as “those who have craving and those who have none”,92 although it is
87 See Burford, 1991, 9. 88 Sn 146. 89 See MW, s.v. trasa. 90 See MW, s.v. sthāvara. 91 tasā, sa-taṇhānaṃ sa-bhayānaṃ c’ etaṃ adhivacanaṃ, tiṭṭhantī ti thāvarā, pahīnataṇhāgamanānaṃ arahantānaṃ etaṃ adhivacanaṃ, Pj I 245 (ad Khp IX.4 = Sn 146). 92 Horner, 1957, 290 (= M II 105).
163 very likely that to the original hearers the meaning was far less restricted, and more on the lines of the meaning in Sanskrit.
There are many other comparable examples of development of meaning. In the Uraga-sutta of the Sutta-nipāta we find the refrain: “That monk leaves behind this shore and the far shore as a snake leaves behind its old skin”.93 What does the phrase “this shore and the far shore” mean in this context? In some Pāli texts “this shore” means “this existence” and “the far shore” means “nibbāna”. By the time the commentator wrote in the fifth century C.E., the idea of leaving behind the far shore in the form of nibbāna was a Mahāyāna idea, which as a Theravādin he was very reluctant to accept. To understand the phrase therefore entails the discussion of the question of whether the Mahāyāna idea could have been in existence at the time when the sutta was composed, and, if not, what “far shore” could mean in this context. My own belief is that the references is not to saṃsāra and to the far shore of saṃsāra, i.e. nibbāna, but to this world and the next, and I believe that the verse was first formulated in a situation where the author was considering this world and the afterlife, rather than the endless stream of saṃsāra. When the Buddhists took the verse over, however, they had to make it fit into the saṃsāra system. In an attempt to avoid the Mahāyāna idea, the commentary on the Sutta-nipāta gives seven explanations,94 including the lower fetters and the higher fetters, and the world of men and the world of gods.95 It is interesting to note that, in the Sanskrit Udāna- varga, where the editions of Chakravarti and Nakatani read orapāraṃ “near and far shore”,96 Bernhard’s edition reads apāraṃ “this shore”, which suggests that the redactor of that version also thought that the abandoning of pāra was unacceptable.
The Buddhaghosuppatti97 states that the commentaries given by Mahinda, and the additions thereto, were burned in a great bonfire when Buddhaghosa had finished writing his commentaries. Although this has long been recognised as being merely an exaggerated way of saying that the Sinhalese commentaries fell into disuse, there is evidence to show that the earlier commentaries were not, in fact, completely superseded. It is clear that the commentaries which Buddhaghosa used to write the historical introduction to the Samantapāsādikā were not destroyed, for they were available to the author of the Mahāvaṃsa and even to the writer of the Mahāvaṃsa-ṭīkā some centuries later.98 The same applies to the commentaries on canonical texts. The Mahāvaṃsa-ṭīkā refers to the Sīhaḷa-aṭṭhakathā on the Majjhima-nikāya and includes information not
93 so bhikkhu jahāti orapāraṃ urago jiṇṇam iva tacaṃ purāṇaṃ, Sn 1–17. 94 Pj II 12–14, discussed by Brough, 1962, 201 foll. 95 oraṃ manussaloko pāraṃ devaloko, Pj II 13, 9. 96 Udāna-v 18.21, etc. 97 Bu-up 7. 98 Malalasekera (1935, cix) dates Mhv-ṭ to the eighth or ninth century
164 given in the Papañcasūdanī.99 The ṭīkā on the Sāratthappakāsinī includes a variant reading for a passage in the Majjhima-nikāya, 100 which is not only not given by Buddhaghosa,101 but which shows by its form ahunā (< adhunā) = idān’ eva that it was taken from a dialect other than Sinhalese Prakrit and Pāli, probably a North Indian Prakrit, and perhaps even from a commentary brought over by Mahinda. Nevertheless, at some unknown date the Sinhalese aṭṭhakathās must have fallen completely into disuse, and they seem now to be irretrievably lost. Only the Pāli commentaries which were based upon them are now extant.
In time, of course, the commentaries themselves became less intelligible, and required explanation. This led to the appearance of ṭīkās, which were in effect commentaries upon the aṭṭhakathās. Sometimes, as I have just said, these commentaries, although late, nevertheless preserved readings and explanations which are superior to anything found in earlier commentaries. In time came a further expansion of commentarial material, with anuṭīkās, and other commentaries being written, and the production of such works has continued right up until the present time, as Buddhist commentators have used all the facilities of Indian philology to explain the meaning of the Buddhavacana, as it has been re-interpreted again and again in the light of ever- changing cultural patterns.
How are we to evaluate the commentarial tradition It is clear that sometimes it is very useful:
(1) The commentaries sometimes explain something we could not otherwise understand. Mrs Rhys Davids, for example, paid tribute to the indispensability of the commentary on the Therīgāthā,102 when she was translating that text. Despite all its errors, it nevertheless gives correct explanations of many passages which would otherwise be quite unintelligible. The same is probably true of almost all commentaries. (2) Commentaries sometimes contain readings which are better than those in the canonical texts we possess. So in the descriptions of famine which occur in the Vinaya and the Saṃyutta-nikāya, there is a word dvīhitikā, which is explained as being made up from du “two” or “difficult” + īhita “effort” or “activity”, i.e. the word means either “there is two-fold effort, i.e. begging will be either successful or unsuccessful” or “life will be difficult”. The commentary upon the Saṃyutta- nikāya, however, gives a variant reading duhitikā, which it
99 Mhv-ṭ 193, 305. 100 M I 255,16 (which is quoted at Spk I 13, 29). 101 Both PTS editions read ahu tañ ñeva aññatarassa. Spk-pṭ (Be 1961) 41, 25 reads ahunā idān’ eva. (Information received from L.S. Cousins in a letter of 25 December 1973). 102 C.A.F. Rhys Davids, 1909, xvi–xvii.
165 says has the same meaning as dvīhitikā. Nevertheless, if duhita is taken as the opposite of suhita “satiated”, then we can see that *duhita would mean “hungry”, and duhitika “stricken by hunger, or connected with hunger”.103 (3) As I have stated, commentaries show us how Buddhist thought has developed since the time when the canonical text they are commenting on was composed.
They were not always quite so useful:
(1) Sometimes the tradition has lost the meaning, and the commentator resorts to giving several explanations, all unlikely, and all very difficult to understand. This produces the familiar cry: “I can understand the text, but not the commentary”. (2) Sometimes the explanations are of a circular nature, e.g. “wise means possessed of wisdom”,104 and are only intelligible if the meaning is already known. (3) Sometimes the explanations are wrong, as I have already pointed out. (4) Sometimes the commentary explanation has had an insidious effect upon the canonical text, i.e. what was originally written in the commentary was sometimes included in the text (as “glosses”), or had an effect upon the words in the text, in that the text was changed to fit the meaning given by the commentary. It has been pointed out105 that the threefold categorisation of kamma (karma), which is found in some Theravādin canonical texts, is due to the misunderstanding of the absolutive upapajja or upapajjaṃ.106 This was thought to be incorrect, and it was consequentially “corrected” to the “locative” upapajje. As a result of this, what had originally been a two-fold classification, i.e. “one feels the result [of a bad deed] in the here and now or, having been reborn, in some future period” became “ … in the here and now, or in (a future) rebirth, or in some future period”. This misinterpretation seems to have come into existence in a 15th-century ṭīkā on the Nettippakaraṇa, from which it was introduced into manuscripts of the Netti itself, and then into manuscripts of the Majjhima-nikāya and the Aṅguttara-nikāya, on which the Netti passage was based.
103 von Hinüber, 1981, 74–86. 104 paṇḍitā ti paṇḍiccena samannāgatā, Sp 552, 24. 105 von Hinüber, 1971, 241–49. 106 The extension of an absolutive by a nasal can be found elsewhere in Pāli. See Geiger, 1994, § 214.
166